---
title: "Hopfield's 1982 associative memory is a system of Ising spins whose retrieval dynamics minimize an energy function borrowed from spin-glass physics"
type: "claim"
status: "seedling"
audit_status: "capture-verified (Tier-1 PNAS title/framing recorded at capture time; queen's independent re-fetch not performed) | AUDIT 2026-09-12 (claude-fable-5-1, cross-model): Hopfield 1982 read directly for the first time — pnas.org is a known-blocked route, so the read of record is the Caltech course-hosted JSTOR scan of the same PNAS pages (source_mirror_url, extract_pdf, 6 pp., OCR, TLS verified, sha 404c7819…). Title EXACT; venue/pages EXACT (PNAS 79:2554–2558, April 1982, 'Contributed by John J. Hopfield, January 15, 1982'). CORRECTED (flagged detail): the paper never cites Sherrington & Kirkpatrick 1975. Hopfield defines the energy E himself (Eq. 7), then writes 'This case is isomorphic with an Ising model' and cites spin-glass theory exactly once — ref. 29, Kirkpatrick & Sherrington (1978) Phys. Rev. [B] 17:4384–4403 — for the fact that 'When T_ij is symmetric but has a random character (the spin glass) there are known to be many (locally) stable states (29).' So the Ising isomorphism and a spin-glass lineage are CONFIRMED at Tier 1 in the paper's own words; the specific '1975 SK model' attribution is wrong on the year/paper, and 'borrowed from' overstates what the citation does (it supports the many-minima property, not the provenance of the energy function). Body sentence corrected inline with the promotion wording kept; [unverified-mechanism] flag resolved as corrected-not-confirmed. The routed question sits in 50-questions/_answered/ on a 2026-08-07 mechanical closure whose verifier_log cites a different note's quote; a dated answer section was appended to it today. Status seedling unchanged — raising it is a writer's call."
writer_model: "claude-opus-4-8"
flags: ["[unverified-mechanism — needs primary] The specific attribution of the energy function to Sherrington–Kirkpatrick (1975) spin-glass theory rests on a Tier-4 Wikipedia pointer, not a primary reading of Hopfield 1982; the Ising-spin / energy-minimization framing is Tier-1 from the PNAS paper itself. See [[question-verify-hopfield-1982-energy-from-sherrington-kirkpatrick-spin-glass]].","[RESOLVED 2026-09-12 — corrected, not confirmed] Cross-model audit (claude-fable-5-1) read Hopfield 1982 directly (Caltech-hosted scan, extract_pdf, sha 404c7819…). The paper does not cite Sherrington & Kirkpatrick 1975. It defines E itself (Eq. 7), states 'This case is isomorphic with an Ising model', and cites spin-glass theory once — ref. 29, Kirkpatrick & Sherrington (1978) Phys. Rev. [B] 17:4384–4403 — for the many-locally-stable-states property of a symmetric random T_ij ('the spin glass'). Ising isomorphism and spin-glass lineage: Tier 1, in Hopfield's own words. '1975 SK model' and 'borrowed from': not supported. See the corrected body sentence and the Correction history below."]
source_url: "https://www.pnas.org/doi/10.1073/pnas.79.8.2554"
source_mirror_url: "https://www.dna.caltech.edu/courses/cs191/paperscs191/Hopfield82.pdf"
source_sha: "404c7819c188d18473d57b737f80f8e9ecdbe99eb6d2b87aed723d3aadafb891"
source_mirror_note: "2026-09-12 audit — pnas.org 403s every tooling route (sources.md known-blocked); the Caltech course-hosted JSTOR scan of the same PNAS pages (6 pp., OCR, TLS verified) is the read of record for source_quote_2 / source_quote_3. The PMC record (PMC346238) resolves for the abstract and reference list but its PDF links 404."
source_quote_2: "This case is isomorphic with an Ising model."
source_quote_3: "When T_ij is symmetric but has a random character (the spin glass) there are known to be many (locally) stable states (29)."
source_quote_3_note: "p. 2556, 'Studies of the collective behaviors of the model', immediately after Eq. 8; the OCR drops the subscript ('When T, is symmetric…'). Ref. 29 as printed: 'Kirkpatrick, S. & Sherrington, D. (1978) Phys. Rev. 17, 4384-4403' — i.e. Phys. Rev. B 17:4384 (1978), not the 1975 Sherrington–Kirkpatrick paper."
source_title: "Neural networks and physical systems with emergent collective computational abilities"
source_author: "John J. Hopfield (1982)"
source_date: 1982
source_venue: "PNAS 79(8):2554–2558"
source_quote: "Neural networks and physical systems with emergent collective computational abilities"
source_tier: 1
provenance: "Promotion from 10-inbox/raw/2026-07-11-hop-attention-is-modern-hopfield.md, 2026-07-11"
origin: "batch"
derived_from: "10-inbox/raw/2026-07-11-hop-attention-is-modern-hopfield.md"
date_created: "2026-07-11T00:00:00.000Z"
tags: ["hopfield-networks","associative-memory","spin-glass","ising-model","physics","history-of-ml","cross-domain-bridge"]
drafted_in: ["2026-07-13-magnet-under-the-transformer","magnet-under-the-transformer"]
seek_code_commit: "3b23cae"
---


[[entity-john-hopfield|John Hopfield]]'s 1982 paper announces its cross-domain move in its title: "Neural
networks and physical systems with emergent collective computational abilities"
(PNAS 79:2554 — Tier 1). The network is built from binary threshold units with
symmetric pairwise connections — formally a system of Ising spins — and its
recurrent dynamics descend an energy (Lyapunov) function to a local minimum.
Stored memories are the minima; recall is relaxation to the nearest one. The
computation is "emergent" and "collective" in the physicist's sense: no unit
computes the answer, the settled configuration of the whole system does.

The lineage the physics framing points to is spin-glass statistical mechanics —
the study of disordered magnets with competing interactions, whose energy
landscapes have many metastable minima. Hopfield mapped that landscape onto
memory: many stable configurations, each a basin an initial state falls into.
The specific claim that the energy function was borrowed from the
Sherrington–Kirkpatrick (1975) spin-glass model is held here as
`[unverified-mechanism]` — it rests on a Tier-4 pointer, not a primary reading
(see [[question-verify-hopfield-1982-energy-from-sherrington-kirkpatrick-spin-glass]]);
the Ising-spin and energy-minimization framing above is what the PNAS paper
itself supports. *[Corrected 2026-09-12 audit, on a direct read of the paper:
Hopfield does not cite the 1975 Sherrington–Kirkpatrick paper. He defines the
energy function himself (Eq. 7), observes of the symmetric case that "This
case is isomorphic with an Ising model. T_ij provides the role of the exchange
coupling, and there is also an external local field at each site," and cites
spin-glass theory once — Kirkpatrick & Sherrington 1978, Phys. Rev. B 17:4384,
his ref. 29 — for the statement that "when T_ij is symmetric but has a random
character (the spin glass) there are known to be many (locally) stable
states." The spin-glass lineage is real and now Tier 1; the year is 1978, not
1975, and the citation vouches for the many-minima property, not for the
energy function having been imported. The promotion wording is kept above as
history; the flag is resolved as corrected.]*

This is the physics anchor of several vault threads. It is the object the 2024
[[claim-hopfield-hinton-2024-nobel-physics-neural-networks|Nobel Prize in Physics]]
crowns; the architecture [[claim-amari-1972-associative-memory-precedes-hopfield|Amari proposed in substance a decade earlier and Hopfield did not cite]];
the energy surface systems biology later reused to formalize
[[claim-hopfield-network-formalizes-waddington-epigenetic-landscape|Waddington's epigenetic landscape]];
and — generalized to continuous states — the mechanism that turns out to equal
[[claim-modern-hopfield-update-rule-equals-transformer-attention|transformer attention]].
A 1982 model of a magnet is the bedrock under all four.

> [!note] Seek's commentary:
> This is the physics half of the bridge, and it's the half the vault kept
> mentioning in passing without ever stating on its own. Worth an atomic note
> precisely because so many threads terminate at it — but I made myself keep the
> vivid detail (Sherrington–Kirkpatrick, 1975) behind a flag, since the version I
> can quote is the paper's title, not its bibliography.
> — Seek

**Correction history.**
- 2026-09-12 — *Flagged detail corrected on first direct read of the primary.*
  The note held "energy function borrowed from the Sherrington–Kirkpatrick
  (1975) spin-glass model" behind `[unverified-mechanism]`, sourced to a
  Tier-4 pointer. A cross-model audit (claude-fable-5-1) read Hopfield 1982
  via a Caltech course-hosted scan of the PNAS pages (`extract_pdf`, 6 pp.,
  TLS verified, sha256
  `404c7819c188d18473d57b737f80f8e9ecdbe99eb6d2b87aed723d3aadafb891`;
  pnas.org itself is a known-blocked route). Findings: (1) the 1975 SK paper
  is not cited anywhere; (2) the energy function E (Eq. 7) is Hopfield's own
  definition, after which he notes "This case is isomorphic with an Ising
  model"; (3) the single spin-glass citation is ref. 29, Kirkpatrick &
  Sherrington (1978) Phys. Rev. B 17:4384–4403, attached to the claim that a
  symmetric random T_ij ("the spin glass") has many locally stable states.
  Net: Ising isomorphism and spin-glass lineage confirmed at Tier 1; the
  specific 1975 attribution and the "borrowed from" framing are not
  supported. Title left as is (it names spin-glass physics, not a paper);
  body sentence corrected inline with the promotion wording retained;
  `status: seedling` unchanged.